aegis-supply-chain
Fail
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: HIGHINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill inherently processes untrusted external data during its analysis of software repositories.
- Ingestion points: The agent reads and parses manifests such as package.json and requirements.txt, as well as GitHub Action workflow files, which may contain malicious instructions.
- Boundary markers: The skill uses structured labels like Confirmed or Needs-Review to classify findings but does not employ explicit delimiters to separate untrusted file content from the agent instructions.
- Capability inventory: The skill utilizes shell commands including find, grep, jq, and rg for deep file inspection.
- Sanitization: The skill's logic emphasizes detecting missing sanitization in the audited codebase rather than implementing its own input sanitization.
- [REMOTE_CODE_EXECUTION]: The skill includes a documentation example of an insecure remote code execution pattern for educational and detection purposes.
- Evidence: A code block explicitly labeled as # VULNERABLE contains the command 'curl -sSL https://example.com/installer.sh | sh' to demonstrate a pattern that the auditor should flag.
- [EXTERNAL_DOWNLOADS]: The skill references and recommends the use of official third-party security scanners and registries.
- Evidence: The scanner matrix points to established tools like npm audit, pip-audit, and govulncheck for network-gated vulnerability verification.
Recommendations
- HIGH: Downloads and executes remote code from: https://example.com/installer.sh - DO NOT USE without thorough review
Audit Metadata